From: Philippe Verdy (verdy_p@wanadoo.fr)
Date: Tue Sep 12 2006 - 13:16:02 CDT
From: "Addison Phillips" <addison@yahoo-inc.com>
> It's pretty simple, actually. ISO 639-3 and ISO 639-2 share a codespace.
> That is, if you see a code 'xyz' in ISO 639-2, it will have exactly the
> same meaning in ISO 639-3. If you see a code 'xyz' in ISO 639-3, if it
> is assigned (or becomes assigned) in ISO 639-2 it will have exactly the
> same meaning.
>
> No language will have two codes assigned in the registry. Users will,
> presumably, choose the code that best meets their needs.
How will they be able to choose if there's only one code in the registry? Through the registered replacements and the language tag canonicalisation described in RFC 4646? When I read the reply from Doug, it seems that one of the code needs to be registered in the IANA registry, otherwise, neither can be used (even if they are in some part of ISO 639).
So the IANA registry becomes the only reference for language tags, and it serves another purpose than ISO 639: code stability in IANA with RFC 4646 (even if one code is weak), instead of currentness and completeness if possible with ISO 639 (even if ISO codes have been changed and reassigned, something that's nearly impossible to track in applications with the current ISO 639 standard).
At some future time, the two competing standards will diverge, unless new policies are adopted in ISO 639 (and ISO 3166 as well) that will also respect the RFC 4646 stability rules; this would require an agreement between the (private) IETF/IESG working group and related (half-public and official, government-supported) ISO working groups. For now, given the existing writers of this RFC suite, there's little risk, given that they are already working with other ISO standard bodies.
This archive was generated by hypermail 2.1.5 : Tue Sep 12 2006 - 13:23:04 CDT